Abnormal brain connectivity in COPD patient correlates cognitive and respiratory impairment

Abstract

Chronic Obstructive Pulmonary Disease(COPD) is a respiratory disease characterized by irreversible airflow limitation and neurological disorders such as cognitive impairment. The prevalence of cognitive impairment among these COPD patients averages 32%, but the analysis of cognitive-related brain functional connectivity in patients has not been thoroughly researched and remains still debatable. Especially, RS fMRI studies focused on respiratory and autonomic functions are uncommon and differ from one to another, so an appropriate brain image analysis pipeline presentation is needed. In this work, we investigated the abnormality of resting-state functional connectivity(RSFC) in the brains of COPD patients, and its relationship with respiratory and cognitive functions. The preprocessing method of the image and the seed points were set differently to compare how the factors of the pipeline affect the final results. Participants were forty-one COPD patients and thirty normal controls (NC). T1-weighted anatomical images and RS fMRI data were acquired for each participant, together with clinical scores for COPD symptoms and cognitive or respiratory function. Group differences between RSFCs across multiple brain regions in COPD patients and NC subjects were obtained by seed-based correlation analysis and the relationship between RSFC and clinical scores was analyzed. Additionally, four distinct seed masks were generated by separating them into denoising or inclusion and spherical or FreeSurfer segmentation, and RSFC map differences between each seed mask were compared. The RSFC maps in the cingulate cortex and occipital-frontal lobe were decreased in the COPD group. In contrast, the COPD groups' RSFC were increased compared to controls in some areas; the caudate/thalamus, temporal lobes, and motor-related regions. The symptoms of COPD and the clinical score related to the patient's quality of life score were positive correlations with motor-related RSFC, and the default mode network areas were a positive correlation with lower connectivity as lung function decreased. When compared to the four image analysis pipelines, In the noise-contained method, a considerable number of clusters have been set as the target region. Also, the noise-removed and spherical mask showed two target regions in one large cluster, but the remaining methods did not. The cingulate cortex is associated with the default mode network, and the decreased connectivity would have affected cognitive function as autonomic dysfunction. The motor-related regions increased RSFC in the COPD group. This appears to be the result of hyperventilation, which causes the chest to move significantly. RSFC results with noise-removed and spherical masks were more accurate and superior than setting seed points to FS masks or including physiological noise in pre-processing.